边生成边思考：在视觉生成过程中交错嵌入文本推理
Thinking-while-Generating: Interleaving Textual Reasoning throughout Visual Generation
把文本推理链交错插入到视觉生成的时间轴上，让模型在生成局部区域时同步进行引导与反思；这一「边生成边推理」的调度方式可直接迁移到自回归图像生成、视频生成以及可控生成的条件注入设计中。
原文依据 textual reasoning is interleaved to both guide upcoming local regions and reflect on previously synthesized ones
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期视觉生成领域的进展越来越多地探索对推理能力的整合。这些工作将文本推理（即「思考」）置于生成过程之前（作为预规划）或之后（作为后精修），但缺乏在生成过程本身之中进行的实时多模态交互。在这项初步研究中，我们提出「边生成边思考」（Thinking-while-Generating, TwiG），这是首个能够在整个视觉生成过程中实现文本推理协同演化的交错式框架。随着视觉内容逐步生成，文本推理被交错插入，既用于引导即将生成的局部区域，也用于对先前已合成的区域进行反思。这种动态交互产生了更具上下文感知能力、语义更丰富的视觉输出。为揭示该框架的潜力，我们考察了三种候选策略：零样本提示、在我们构建的 TwiG-50K 数据集上进行监督微调（SFT），以及通过定制的 TwiG-GRPO 策略进行强化学习（RL），每种策略都为交错推理的动态特性提供了独特洞见。我们希望本工作能启发更多关于交错文本推理以增强视觉生成的研究。代码已发布于：https://github.com/ZiyuGuo99/Thinking-while-Generating。
英文原文摘要 1261 字符
Recent advances in visual generation have increasingly explored the integration of reasoning capabilities. They incorporate textual reasoning, i.e., think, either before (as pre-planning) or after (as post-refinement) the generation process, yet they lack on-the-fly multimodal interaction during the generation itself. In this preliminary study, we introduce Thinking-while-Generating (TwiG), the first interleaved framework that enables co-evolving textual reasoning throughout the visual generation process. As visual content is progressively generating, textual reasoning is interleaved to both guide upcoming local regions and reflect on previously synthesized ones. This dynamic interplay produces more context-aware and semantically rich visual outputs. To unveil the potential of this framework, we investigate three candidate strategies, zero-shot prompting, supervised fine-tuning (SFT) on our curated TwiG-50K dataset, and reinforcement learning (RL) via a customized TwiG-GRPO strategy, each offering unique insights into the dynamics of interleaved reasoning. We hope this work inspires further research into interleaving textual reasoning for enhanced visual generation. Code is released at: https://github.com/ZiyuGuo99/Thinking-while-Generating.